Say what you will about transhumanism, but one thing’s for certain: it really polarises opinion, and nowhere more so than in the halls of academia and scientific research. Observe: Wired/Gizmodo had a chat with Singularitarian-in-chief Ray Kurzweil, who restated his theory (considered unrealistically optimistic by some transhumanists) that we’ll be able to reverse-engineer the human brain and simulate it with computers within a decade or so.
Here’s how that math works, Kurzweil explains: The design of the brain is in the genome. The human genome has three billion base pairs or six billion bits, which is about 800 million bytes before compression, he says. Eliminating redundancies and applying loss-less compression, that information can be compressed into about 50 million bytes, according to Kurzweil.
About half of that is the brain, which comes down to 25 million bytes, or a million lines of code.
Now enter PZ Myers, prominent atheism advocate (I like to think of him as “Dawkins’ Bulldog”, though I’m not sure Dawkins really needs a bulldog in the way that Darwin did) and vigorous debunker of fringe science. Broad claims in the Kurzweil vein are like a red rag to Myers, especially on his home turf of genetic biology, and he’s not afraid of mixing in a little ad hominem disparagement with his rejoinders, either:
Kurzweil knows nothing about how the brain works. It’s design is not encoded in the genome: what’s in the genome is a collection of molecular tools wrapped up in bits of conditional logic, the regulatory part of the genome, that makes cells responsive to interactions with a complex environment. The brain unfolds during development, by means of essential cell:cell interactions, of which we understand only a tiny fraction. The end result is a brain that is much, much more than simply the sum of the nucleotides that encode a few thousand proteins. He has to simulate all of development from his codebase in order to generate a brain simulator, and he isn’t even aware of the magnitude of that problem.
[…]
To simplify it so a computer science guy can get it, Kurzweil has everything completely wrong. The genome is not the program; it’s the data. The program is the ontogeny of the organism, which is an emergent property of interactions between the regulatory components of the genome and the environment, which uses that data to build species-specific properties of the organism. He doesn’t even comprehend the nature of the problem, and here he is pontificating on magic solutions completely free of facts and reason.
Now, I’m not taking sides here*; I don’t know enough computer science or evolutionary biology to cut into either interpretation. But a high-minded slapfight like this is always of interest, because it highlights just how seriously some very intelligent people take the issue. Kurzweil has more than a tinge of the evangelist about him, which is (I suspect) a large part of what bothers Myers about him, but there’s obviously something powerful about the idea (the meme?) of transhumanism/singularitarianism that he feels makes it worth fighting.
Ideas that get people arguing are important ideas. I consider myself a fellow traveller of transhumanism for this very reason; the ways we imagine tomorrow says a lot about where we are today, and vice versa. There’s a lot to learn by listening to both sides, I think.
[ * Yeah, yeah, I know, I’ve got marks on my ass from sitting on the fence. That’s just how I roll, baby; you want clenched-fist advocacy of anything but the right to think for yourself, you’re gonna need to read a different blog. ]
But a high-minded slapfight like this is always of interest, because it highlights just how seriously some very intelligent people take the issue.
Wait. What?
Are you implying that PZ takes the issues he argues about seriously?
I would rather say that most of his rants are on what he considers serious problems, seriously deluded liars, or cases of serious insanity. Which is quite a different thing.
There are many intelligent people who take transhumanism seriously. But to take PZ calling Kurzweil a crook who “is pontificating on magic solutions completely free of facts and reason”, as a sign for that, is a little bit of a stretch.
If you go back into PZ’s archives, this isn’t the first time he has demolished Kurzweil. I remember one where Kurzweil was hyping his brain graph and Moore’s law to show the singularity would happen. Myer’s then did a graph of brains and showed how Kurzweil was cherry picking and even lying about data points in the extreme to show that there is no consistent curve.
I think Myers goes off the deep end, Unfortunately, many of his points are correct.
Kurzweil and the “geek rapture” fail to comprehend the sheer unadulterated complexity of the “mind”. Even the strongest materialists understand that.
Just go back 40 years and see how soon futurists thought true AI was. Today, we have nothing close to machine based consciousness as we humans know it. Hell, we really don’t have even a strong scientifically based definition of consciousness. It’s a lot like what Justice Brennan said about pornography, “..when we see it, we know it”.
In other words, I am not saying definitively that the singularity won’t happen. I am saying that it’s along way away and, it will npt be a singularity. Progress in fields like genetics, nanotech and computer science will proceed. And there will be changes to the human archetype. But, there will probably not be the Vinge-Kurzweil singularity, We’ll look back at some point and realize that wow!, we sure have changed.
The same could be said about humanity 500 years ago.
Uncertainty FTW!
I agree with Myers that the brain is far too complex for us to comprehend in a program anytime soon. But I think a bigger question is how much we can accomplish soon enough for it to make a big enough difference in our lives that we would call it a Singularity. and I agree with Rick York that we might only appreciate the change in retrospect.
I find it strange that it’s always the most stalwartly self-identified atheists who most stridently argue that there’s something Special and Ineffable about human consciousness that can never be replicated artificially.
I have always had a problem with Kurzweil’s data cherry picking, equivocation and orthogonal arguments for imminent nerd rapture.
His original argument way back in the cyber-pleistocene involved a syllogism like,
1.) The human brain computes at 20 quadrillion FLOPS per second
2.) Moore’s Law says we’re increasing computer processing power at an exponential rate such that we’ll have 20 quadrillion + FLOP computers in 40 years.
3.) We’ll duplicate the brain in a computer in 40 years.
On the surface, it might sound reasonable to some. Until you realize that when it comes to human intelligence and consciousness, it is not speed that is the problem. You could have a googol-plex FLOP computer that does septillion arithmetic instructions, updates a gajillion Facebook pages, renders a quintillion zombie heads blowing up simultaneously in 2080p HD on a sesquipedillion different screens simultaneously every second. That impossibly fast computer is still stupider than the schlock it rode in on. You need the software, duh. And there is no clear exponential graph leading up to when we figure out the software for brains.
Nevermind the mind-meltingly complex problems of how to calculate actual brain “computing power” let alone the question of whether the human mind is replicable in a computer.
Enter exhibit B: the human brain as a subset of the human genome (!??!??!?!).
Forget that the brain is a fully expressed phenotype, the ontogeny of interactions argument, the embodiment problem, and all the other associated issues.
Kurzweil is running ultimately the same bait-n-switch orthogonal argument from before in different clothes. “The brain is (quantity x), we can do things in computers at (quantity x) level, therefore, brain is going to happen soon (like before I die)”. TFLOPS, bytes, lines of code, genome sequences. These are irrelevant to whether we’ll replicate human or human-level minds in computers. They are probably necessary, but not sufficient. What is necessary is the *understanding* – the concepts, paradigms, perspectives, those messy, unpredictable breakthroughs, those Black Swans, highly entwined with human imagination.
E=MC^2. That’s encodable in about 6 bytes. Where is the “exponential graph” that tells us when that theory is conceptualized? There is none. Sure, there are more scientists and innovators and people actively pursuing the Singularity now. Is that empirical evidence that we are going to make all those necessary conceptual breakthroughs as well as the resources and socio-economic circumstances to have brains in computers in 20 years? No.
We might call these Singularity arguments “Argumentum ad admiratio” – argument from sense-of-wonder, or argument from hype*. “Look at all this cool amazing stuff we’re creating, and look how it’s all exponentially accelerating! How can you *not* believe we’ll achieve (blank) in X years?!!”
Doing lots of cool impressive stuff really fast now does not mean we’ll make the tough, unpredictable breakthroughs necessary to do some other really cool thing in the future.
(*Sometimes known as ‘the lame tech conference sales pitch’)
Can anyone explain why we need to make exact copies of brains in a computer? There’s 6 or 7 billion humans with varying amounts of brain already. Why not make the computer do something outside the capabilities of a neuron-based brain (like thinking of an appropriate use for AI)?
Have you hear about the Bluebrain project? It seems it’s not just Ray Kurzweil who believes that brain can be reverse-engineered:
http://bluebrain.epfl.ch/
http://www.ted.com/talks/henry_markram_supercomputing_the_brain_s_secrets.html
I’m not a rocket surgeon or anything, but as I understand it, Kurzweil isnt suggesting that we can replicate a brain simply by mapping the human genome. I thought the idea was to effectively replace 1 neuron with a simulated neuron at a time, and then, once they were all replaced, we’d have a brain that didnt know it had been completely replaced. That way we leave the “software” intact and dont care about it, and while that procedure is still a long way off, its much easier to conceptualize as a single neuron and its functions are a lot more understood than the entire brain and its development. Clearly its got to be more complex, there’s several kinds of neurons and bits and pieces in there but with nano tech on the cusp, it doesnt seem too unreasonable to see that happening in 30-50 years.
It’s worth noting that the Blue Brain Project- a multibillion dollar projected funded by the Swiss government, with the intention of simluating a human brain, is already underway and expects to finish by approximately 2019, which is 10 years *before* Kurzweil predicts:
http://news.bbc.co.uk/2/hi/8164060.stm
This is, to my knowledge, the only time a credible source has
The problem with PZ’s argument is that assumes a successful simulation would need to know everything about the brain it was simulating. Current research is instead aiming to simply take a sufficiently high-resolution scan of a real brain, put it in a computer, and operate it, without knowing what each part of the simulated brain is for. With this approach, the only thing we need to really understand is sensory information going in and out of the brain stem, which is a surmountable task.